P21 Community v4.0.1 no longer checks labels (SD0063 and SD0063A)

Sergiy

on September 29, 2022

P21

CDISC removed these rules for SDTMIG 3.3 and higher versions.

Hi Jim, 

Please confirm the version of your SDTM-IG?

SD0063/A rules should work for all SDTM-IG up to 3.2. CDISC removed these rules for 3.3 and higher versions.

Kind Regards,
Sergiy

Sorry...v3.3.  Any idea why…

Sorry...v3.3.  Any idea why were these checks were removed? 

Labels

To extend on Sergiy's answer and some additional explanation:

In the past, there was some critique on the conformance requirement that the labels on SDTM variables must be exactly as in the IG, or, when takenfrom the SDTM model, exactly as from the model, and all this, case sensitive.
Labels are information to the reviewer what the variable is about, and thus may depend on the context. Having to have them exactly as in the IG or model does not allow the sponsor to state what the variable is about in their own submission context, i.e. for their own study case.
Furthermore, the limitation to 40 (ASCII-only) characters, due to the outdated SAS Transport format, led to a lot of confusion, with "artificial" labels which needed abbreviations to comply to the 40-character limitation. This led to a lot of discussions about the correct implementation in validation software, also as even CDISC has published labels > 40 characters in the past.

To allow sponsors to better explain what the variable means, this requirement was dropped as of SDTMIG v.3.3. On page 423 (Appendix E) it states in the "Revision History": "Removed variable labels from conformance criteria".
This means that the sponsor is now allowed to change the label when it feels that it doesn't explain well what the variable is about. For example, in a Corona Virus study, the sponsor may feel that for MBCAT ("Category"), the label "Virus Test Category" is more appropriate, and may then use that.

Having dropped the requirement is a very good thing, as the labels can now be used for what they are meant: to explain what the variable is about in the context of the specific study.

Thanks, Josef, for the…

Thanks, Josef, for the explanation. Not sure I agree with the change, but there's lots of stuff I disagree with, so what's one more!

Sergiy

on October 6, 2022

P21

Hi Jim,  My impression is…

Hi Jim, 

My impression is that CDISC removed the rule for standard variable labels to avoid industry criticism for publishing invalid or incorrect labels in SDTM-IG.

Sometimes CDISC team could not handle 40-chars limitation and had other errors in variable labels most likely due to lack of resources for QC process. In general, it's challenging to avoid any errors in 300 page documentation. In many cases, errors can be identified only during practical implementation of specs. So, now CDISC says that standard variable labels are not important, and users can even replace them with custom labels.

Yes, it's embarrassing for standard development organization to make such trivial errors like exceeding a limit of 40-chars in variable label. It can be easily caught in Excel or many other tools. Even more easy solution for CDISC is to claim that this is not important and should be ignored.

My concern is that when when users start ignoring standards labels, the next steps are changing labels and then changing interpretation or misuse of standard variables. In addition to SD0063 rule, we have SD0063A rule which looks to so-called Model Permissible variables, which are not present in SDTM-IG but may be added to standard domains for SDTM Model. SD0063A rule use variable labels from SDTM Model. From my experience, in 10% or reported validation issues are real problems about incorrect usages of additional standard variables from SDTM Model.

Our Enterprise users can configure validation rules for their needs. For example, adding SD0063 rule to SDTM-IG 3.3.

Community version of validator represents official rules from CDISC, FDA, PMDA. 

Kind Regards,
Sergiy

"My concern is that when…

"My concern is that when when users start ignoring standards labels, the next steps are changing labels and then changing interpretation or misuse of standard variables."

I agree 100%.